Connector reliability test method integrating artificial intelligence and complex environment simulation

Abstract

The invention relates to the technical field of model simulation, in particular to a connector reliability test method fusing artificial intelligence and complex environment simulation, and is used for solving the technical problem that the evaluation result of the connector reliability is not accurate enough. The method comprises the following steps: acquiring various working parameter data and life data of a plurality of connectors in an accelerated life test process; determining the basic relevancy of each type of working parameters; constructing a parameter influence chain based on the influence transfer relationship among the multiple types of working parameters; determining the transmission weight of each type of working parameters in the parameter influence chain, wherein the transmission weight is used for representing the indirect influence degree of the working parameters influenced by the superior working parameters on the service life; determining an influence coefficient of each type of working parameters based on the basic relevancy and the transmission weight of each type of working parameters; and based on the influence coefficient, completing model simulation through data fusion simulation and artificial intelligence model training, and constructing a connector reliability evaluation model.

Description

Technical Field

[0001] This application relates to the field of model simulation technology, specifically to a connector reliability testing method that integrates artificial intelligence and complex environment simulation. Background Technology

[0002] Electrical connectors are commonly used in mechanical equipment to transmit power or signals, and their reliability directly affects the stable operation of the entire system. Currently, connector reliability testing primarily relies on a combination of accelerated life testing and simulation analysis. This involves applying stresses higher than normal to shorten the testing cycle and using simulation models to model the connector's behavior under multi-physics coupling environments. However, this approach struggles to comprehensively and accurately characterize the combined impact of various environmental and operational parameters on reliability when faced with complex operating conditions involving multiple interacting parameters, leading to inaccurate reliability assessments of the connectors. Summary of the Invention

[0003] To address the technical problem of inaccurate connector reliability assessment results, this application aims to provide a connector reliability testing method that integrates artificial intelligence and complex environment simulation. The specific technical solution adopted is as follows: Acquire various operating parameter data and lifespan data of multiple connectors during accelerated life testing; Determine the basic correlation degree for each type of working parameter. The basic correlation degree is used to characterize the degree of direct correlation between working parameters and lifespan. Construct parameter influence chains based on the influence transmission relationships among multiple types of working parameters; Determine the propagation weight of each type of working parameter in the parameter influence chain. The propagation weight is used to characterize the degree of indirect influence of the working parameter on the lifespan due to the influence of the superior working parameter. Based on the basic relevance and transit weight of each type of working parameter, the influence coefficient of each type of working parameter is determined; Based on the influence coefficient, a connector reliability assessment model is constructed through data fusion simulation and artificial intelligence model training. The connector reliability assessment model is used to predict the life of the connector based on the influence coefficient of the connector.

[0004] In one possible implementation, determining the basic correlation of each type of operating parameter includes: calculating the out-of-class coefficient for each connector under the first operating parameter, whereby the out-of-class coefficient is used to characterize the overall similarity between the corresponding connector and other connectors under the second operating parameter; the second operating parameter is the operating parameter other than the first operating parameter; using the out-of-class coefficient of each connector as the weight, performing a weighted linear fit between the first operating parameter and the lifetime data; and using the absolute value of the slope of the fitted line as the basic correlation of the first operating parameter.

[0005] In one possible implementation, calculating the out-of-class coefficient for each connector under the target operating parameters includes: calculating the difference between the target connector and any other connector on all second operating parameters as the out-of-class distance, wherein the difference is the ratio of the absolute value of the difference between the target connector and any other connector on all second operating parameters to the sample mean of the second operating parameters; taking the mean of all out-of-class distances of the target connector as the overall degree of difference of the target connector; and determining the out-of-class coefficient based on the overall degree of difference, wherein the out-of-class coefficient is inversely proportional to the overall degree of difference.

[0006] In one possible implementation, determining the transfer weight of each type of working parameter in the parameter influence chain includes: determining the superior working parameter of the target working parameter in the parameter influence chain; calculating the superior difference coefficient between any two connectors based on the difference in real-time monitoring values ​​of different connectors under the superior working parameter and the out-of-class distance; determining the superior transfer coefficient based on the superior difference coefficient, and the deviation of the connector's real-time monitoring value on the target working parameter from the standard value; and determining the transfer weight of the target working parameter based on the superior transfer coefficients of multiple connectors and the difference in the rate of change of performance degradation rate before and after the transfer moment.

[0007] In one possible implementation, the superior difference coefficient of the connectors is calculated based on the difference in real-time monitoring values ​​of different connectors under the superior operating parameters and the out-of-class distance. This includes: determining the absolute value of the difference between the real-time monitoring values ​​of the first connector and the second connector under the superior operating parameters; the first connector is any one of a plurality of connectors, and the second connector is any one of a plurality of connectors other than the first connector; calculating the out-of-class distance of the first connector and the second connector under the superior operating parameters; and using the ratio of the absolute value to the out-of-class distance as the superior difference coefficient of the first connector and the second connector under the superior operating parameters.

[0008] In one possible implementation, the superior transmission coefficient is the product of the superior difference coefficient and the first absolute value, where the first absolute value is the absolute value of the difference between the real-time monitored value of the target working parameter and the standard value.

[0009] In one possible implementation, the transfer weight of the target operating parameter is determined based on the upper-level transfer coefficients of multiple connectors and the difference in the rate of change of performance degradation before and after the transfer moment. This includes: calculating the upper-level transfer coefficients between the first connector and each of the second connectors, and determining the maximum upper-level transfer coefficient; determining the moment corresponding to the maximum upper-level transfer coefficient as the transfer moment of the first connector; calculating the average rate of change of the first connector after the transfer moment and the first difference between the average rate of change of the first connector and the average rate of change of the first connector before the transfer moment; for each connector, calculating the product of the first difference and the corresponding maximum upper-level transfer coefficient; and averaging all the products and then performing linear normalization to obtain the transfer weight of the target operating parameter.

[0010] In one possible implementation, the influence coefficient of each type of working parameter is determined based on the basic relevance and transfer weight of each type of working parameter. This includes: integrating and calculating the basic relevance of the target working parameter, the transfer weight of the target working parameter, and the transfer weight of the subordinate working parameters of the target working parameter, and normalizing them to obtain the influence coefficient of the target working parameter. The influence coefficient is positively correlated with the basic relevance of the target working parameter and the transfer weight of the subordinate working parameters, and negatively correlated with the transfer weight of the target working parameter.

[0011] In one possible implementation, a connector reliability assessment model is constructed based on the influence coefficient through data fusion simulation and artificial intelligence model training. This includes: establishing a parameterized finite element model using operating parameters with influence coefficients higher than a preset threshold as input variables; conducting virtual accelerated life tests in a simulation environment to generate a simulation dataset including parameter evolution and virtual life; inputting the simulation dataset into a data generation model to generate virtual sample data; the data generation model is a pre-trained neural network model for data generation; and generating a training set based on actual test data, simulation dataset, and virtual sample data to train the physical information neural network, thereby obtaining the connector reliability assessment model.

[0012] In one possible implementation, multiple types of operating parameter data and lifetime data of multiple connectors are acquired during accelerated life testing, including: randomly assigning multiple connectors to different stress groups for accelerated life testing, the stress groups including temperature stress group and humidity stress group; multiple types of operating parameters including temperature parameters and humidity parameters; periodically monitoring the environmental parameters and insulation resistance of each connector during the test; and determining the lifetime data of each connector based on the time it takes for the insulation resistance to drop to a preset percentage of the initial value.

[0013] This application offers the following advantages: It determines the influence weight of each type of operating parameter by using a basic correlation coefficient to characterize the direct correlation between operating parameters and lifetime, and a transitive weight to characterize the indirect influence of operating parameters on lifetime. Based on this comprehensive influence weight, a reliability assessment model integrating simulation and artificial intelligence is constructed. Therefore, this application considers not only the direct relationship between operating parameters and lifetime but also the indirect relationship, enabling the final assessment model to uncover both the direct and indirect relationships between operating parameters and lifetime, thus making the assessment results more accurate. Attached Figure Description

[0014] To more clearly illustrate the technical solutions and advantages in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0015] Figure 1 A flowchart illustrating a connector reliability testing method integrating artificial intelligence and complex environment simulation, provided as an embodiment of this application. Figure 1 ; Figure 2 A flowchart illustrating a connector reliability testing method integrating artificial intelligence and complex environment simulation, provided as an embodiment of this application. Figure 2 ; Figure 3 A flowchart illustrating a connector reliability testing method integrating artificial intelligence and complex environment simulation, provided as an embodiment of this application. Figure 3 ; Figure 4 A flowchart illustrating a connector reliability testing method integrating artificial intelligence and complex environment simulation, provided as an embodiment of this application. Figure 4 . Detailed Implementation

[0016] To further illustrate the technical means and effects adopted by this application to achieve the intended inventive objective, the following, in conjunction with the accompanying drawings and preferred embodiments, details the specific implementation, structure, features, and effects of a connector reliability testing method integrating artificial intelligence and complex environment simulation proposed in this application. In the following description, different "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. Furthermore, specific features, structures, or characteristics in one or more embodiments can be combined in any suitable form.

[0017] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application pertains.

[0018] The following description, in conjunction with the accompanying drawings, details a specific scheme for a connector reliability testing method that integrates artificial intelligence and complex environment simulation, as provided in this application.

[0019] Please see Figure 1 The diagram illustrates a flowchart of a connector reliability testing method integrating artificial intelligence and complex environment simulation, provided in one embodiment of this application. The method includes: Step 101: Obtain various operating parameter data and life data of multiple connectors during accelerated life testing.

[0020] As one possible implementation, this step involves randomly assigning multiple connectors to different stress groups for accelerated life testing, including temperature stress groups and humidity stress groups; multiple operating parameters including temperature parameters and humidity parameters; during the test, the environmental parameters and insulation resistance of each connector are periodically monitored; and the life data of each connector is determined based on the time it takes for the insulation resistance to drop to a preset percentage of its initial value.

[0021] Specifically, a number of electrical connectors are randomly selected as test connectors (e.g., 45). Test groups with different stress conditions are set based on the Arrennis acceleration model, including temperature stress group, humidity stress group, and other stress conditions (corresponding to various operating parameters). Connectors are randomly assigned to each stress group and run under the set conditions. During the test, environmental parameters such as temperature and humidity of each connector are collected at fixed time intervals (e.g., every 2 hours), and their insulation resistance value is monitored to characterize the reliability status. The failure threshold is determined by using a decrease in insulation resistance to a certain percentage (e.g., 10%) of the initial value, thereby determining the lifespan data of each connector.

[0022] Step 102: Determine the basic correlation of each type of working parameter.

[0023] Among them, the basic correlation is used to characterize the degree of direct correlation between working parameters and lifespan.

[0024] As one possible implementation, this application quantifies the overall differences among different connectors in parameters other than the target parameter, evaluates the credibility weight of each sample data in the analysis of the target parameter, and then performs a weighted analysis on the relationship between the parameter and lifetime based on the weight, and finally extracts an index that can accurately characterize the direct correlation strength between such parameter and lifetime.

[0025] Understandably, connectors face complex environmental influences in actual operation, and changes in parameters such as temperature and humidity directly affect their working condition and reliability. For example, increased temperature may cause material expansion and deformation, thereby affecting connection stability. This step, based on connector test results, analyzes the correlation between changes in various parameters and connector lifespan to determine the basic correlation of each type of parameter, thereby reflecting the trend of parameter changes' impact on lifespan.

[0026] Based on the test environment parameters and corresponding lifespan results for each connector, a preliminary assessment of the correlation between various parameters and product lifespan can be made. By comparing the lifespan differences of connectors under different test environments, the impact of environmental parameter changes on lifespan can be analyzed. This step establishes a correlation model between various parameters and connector lifespan by examining the relationship between the differences in test environments for different connectors and their final lifespan performance, thus characterizing the impact of each parameter on reliability. By analyzing the correlation between the distribution trend of all connectors on a certain type of parameter and its lifespan change trend, the basic correlation of that type of parameter can be obtained. During the analysis, the smaller the differences in other types of parameters, i.e., when other variables are well controlled, the higher the reliability of the observed trend. Specifically, the multi-dimensional parameter data of all connectors are projected onto a single parameter dimension to study the correlation between the distribution of connectors on that parameter dimension and their lifespan. The higher the consistency of the connector distribution in the projection direction, the higher the reliability of the observed trend.

[0027] Step 103: Construct parameter influence chains based on the influence transmission relationship between multiple types of working parameters.

[0028] As one possible implementation, this application pre-establishes a cascade action path from the initial parameter to the final influencing parameter based on the known physical interaction mechanism between the working parameters of the connector in a multi-physics coupling environment. This forms a structured chain model that can fully describe the unidirectional influence transmission relationship between parameters, providing a theoretical basis and relational framework for subsequent quantification of the indirect influence between parameters.

[0029] For example, during the operation of an electrical connector, the current flowing inside generates Joule heating, causing the material to expand and altering the stress state of the seal. This change in stress state further affects the magnitude of the sealing contact pressure, which in turn affects the degree of moisture or water vapor penetration. Increased moisture penetration triggers electrochemical corrosion, ultimately affecting the normal operating performance of the electrical connector. Therefore, in the operating environment of an electrical connector, there exists a strong coupling relationship between the parameters of "electricity-heat-pressure-humidity," meaning that there is a clear cascading influence mechanism between different physical field parameters. A change in one parameter condition will indirectly cause changes in other parameters through this influence chain, thus having a comprehensive impact on the reliability of the connector. Based on the known relationships between parameters, a parameter influence chain reflecting the transmission path of influence between parameters can be constructed to intuitively characterize the causal transmission relationship between multiple parameters.

[0030] As an example, the parameter influence chain constructed in this application is "electricity-heat-pressure-humidity", where electricity is the first-level working parameter, heat is the second-level working parameter, pressure is the third-level working parameter, and humidity is the fourth-level working parameter.

[0031] Step 104: Determine the transfer weight of each type of working parameter in the parameter influence chain.

[0032] Among them, the transitive weight is used to characterize the degree of indirect impact of the working parameters on the lifespan caused by the influence of the superior working parameters.

[0033] As one possible implementation, this application establishes a correlation model between the indirect influence between parameters and the change in reliability status by quantitatively analyzing the degree of deviation of the actual state of the lower-level parameters caused by the difference of the upper-level parameters, and combining the dynamic of connector performance degradation when the deviation occurs. This systematically characterizes the degree of indirect influence of various working parameters on the upper-level parameters in the influence chain.

[0034] Specifically, in the parameter influence chain, different parameters are cascaded and transmitted along a pre-defined path: the closer a parameter is to the beginning of the influence chain, the less it is affected by the transmission from other parameters, while the transmission effect of changes in that parameter itself on downstream parameters is more significant. Taking any type of parameter as an example, by analyzing the correlation between the differences in this parameter among different samples and their corresponding lifetime differences, we can reveal the comprehensive impact of the parameter's changes on the connector lifetime through the indirect effects of the influence chain, and thus obtain its transmission weight.

[0035] Step 105: Determine the influence coefficient of each type of working parameter based on the basic correlation and transmission weight of each type of working parameter.

[0036] As one possible implementation, this application normalizes and integrates the basic correlation of the direct influence of the characterization parameter with the transmission weight of the indirect influence of the characterization parameter to form a comprehensive evaluation index that can simultaneously reflect the role of the parameter itself and its pivotal position in the influence chain, thereby achieving a comprehensive quantification of the degree of influence of the reliability of various working parameters.

[0037] Step 106: Based on the influence coefficient, construct a connector reliability assessment model through data fusion simulation and artificial intelligence model training. The connector reliability assessment model is used to predict the life of the connector based on the influence coefficient of the connector.

[0038] As one possible implementation, this step can be specifically implemented as follows: using working parameters with influence coefficients higher than a preset threshold as input variables, a parameterized finite element model is established; virtual accelerated life testing is performed in a simulation environment to generate a simulation dataset including parameter evolution and virtual life; the simulation dataset is input into a data generation model to generate virtual sample data; the data generation model is a pre-trained neural network model for data generation; a training set is generated based on actual test data, simulation dataset, and virtual sample data, and the physical information neural network is trained to obtain a connector reliability assessment model.

[0039] Specifically, this step can be implemented as follows: obtain the evaluation results of the influence coefficients of all parameters; select parameters with influence coefficients higher than a set threshold (e.g., greater than 0.5) as key input variables, and establish a parameterized finite element model.

[0040] The system performs a virtual accelerated lifetime test in a simulation environment, adjusts the values ​​of the key input variables, and generates an initial simulation dataset that includes the parameter evolution process and the corresponding virtual lifetime.

[0041] The initial simulation dataset is augmented using neural network technology to generate an extended virtual sample dataset, thereby significantly improving the parameter coverage and operating condition diversity of the dataset.

[0042] A comprehensive reliability evaluation model is constructed by training a physical information neural network based on a hybrid dataset consisting of real test data, an initial simulation dataset, and an extended virtual sample dataset. This model takes the key input variables as input and directly outputs the connector's life prediction value and failure probability assessment result.

[0043] Based on the above technical solution, this application determines the influence weight of each type of operating parameter by using a basic correlation coefficient to characterize the direct correlation between operating parameters and lifetime, and a transitivity weight to characterize the indirect influence of operating parameters on lifetime. Then, based on this comprehensive influence weight, a reliability assessment model integrating simulation and artificial intelligence is constructed. Therefore, this application considers not only the direct relationship between operating parameters and lifetime but also the indirect relationship, enabling the final assessment model to uncover both the direct and indirect relationships between operating parameters and lifetime, thus making the assessment results more accurate.

[0044] like Figure 2 As shown, in one possible implementation, the process of determining the basic relevance of each type of working parameter in step 102 above can be specifically implemented through the following steps: Step 201: Calculate the out-of-class coefficient for each connector under the first operating parameters.

[0045] The out-of-class coefficient is used to characterize the overall similarity between the corresponding connector and other connectors in the second operating parameter; the second operating parameter is the operating parameter other than the first operating parameter.

[0046] As one possible implementation, this step can be specifically implemented as follows: calculate the difference between the target connector and any other connector on all second operating parameters as the out-of-class distance, where the difference is the ratio of the absolute value of the difference between the target connector and any other connector on all second operating parameters to the sample mean of the second operating parameters; take the mean of all out-of-class distances of the target connector as the overall degree of difference of the target connector; and determine the out-of-class coefficient based on the overall degree of difference, where the out-of-class coefficient is inversely proportional to the overall degree of difference.

[0047] Optionally, taking the m-th parameter as an example, the out-of-class distance is obtained by calculating the differences of all connectors on parameters other than the m-th parameter, in order to evaluate the dispersion of the projected distance of the connectors on the m-th parameter dimension.

[0048] As an example, the out-of-class distance between the k-th connector and the k'-th connector in the m-th class parameter. Satisfy the following formula: in, This indicates the total number of categories of the parameter. The value of is greater than or equal to 2. Indicates non-first Other parameters of the class parameter, Indicates the first The connector in the first The value of the class parameter here is taken from the standard value set by the corresponding connector during testing. Indicates the first The connector in the first The value of the class parameter, Indicates that all connectors are in the first... The mean of the class parameters. Indicates the first The connector and the first The differences in these parameters among the connectors, The denominator is used to balance the excessive numerical differences caused by the differences in the magnitudes of different types of parameters, and then the ... The connector and the first The connector is in all except the first The mean of the differences in all parameters outside the class parameter is used as the out-of-class distance between the two connectors.

[0049] It should be noted that, to ensure the calculation results are meaningful, in this embodiment of the application, when performing fractional operations, if the denominator is 0, a parameter adjustment factor can be added to the denominator for addition to prevent the denominator from being 0. This parameter adjustment factor is a very small positive number. For example, the value of this parameter adjustment factor can be 0.01. Its specific value can be set by the implementer according to the actual situation, and this embodiment of the application does not make a specific limitation. Furthermore, whether the parameter adjustment factor has dimensions can change with the parameter or feature it is added to. Specifically, if the parameter or feature it is added to is dimensionless, then the parameter adjustment factor is dimensionless; if the parameter or feature it is added to has dimensions, then the dimension of the parameter adjustment factor is the same as the dimension of the parameter or feature.

[0050] Based on the above method, the first two connectors can be obtained at the... After calculating the out-of-class distance on the class parameter, the first... The average distance between each connector and all other connectors outside the same class, and then applying an exponential function. The connector was obtained in the first Out-of-class coefficients on class parameters The out-of-class coefficient is negatively correlated with the average out-of-class distance; the larger the out-of-class coefficient, the greater the out-of-class distance for the connector, excluding the first class. The more similar the parameter settings (excluding class parameters) are to those of other connectors, the better the analysis of the first... The higher the reliability of the connector data, the more relevant the class parameters are to the connector's lifespan.

[0051] Step 202: Using the out-of-class coefficients of each connector as weights, perform a weighted linear fit on the first operating parameter and the lifetime data.

[0052] Step 203: Use the absolute value of the slope of the fitted line as the basic correlation of the first working parameter.

[0053] Optionally, after determining the out-of-class coefficients, a weighted linear fit is performed with each class of parameter values ​​as the independent variable and lifetime as the dependent variable, using the corresponding out-of-class coefficients as weights. The absolute value of the slope of the fitted line is defined as the basic correlation of that parameter. A higher basic correlation indicates a more significant impact of the parameter's change on the connector's lifetime.

[0054] Based on the above technical solution, this application determines the basic correlation by introducing out-of-class coefficients as weights for weighted linear fitting. This method effectively reduces the interference caused by differences in other parameters when analyzing the relationship between parameters and lifetime, thereby more accurately characterizing the direct correlation between working parameters and lifetime, avoiding the bias caused by parameter cross-effects in traditional correlation analysis, and improving the accuracy and reliability of direct effect analysis.

[0055] like Figure 3 As shown, in one possible implementation, the process of determining the transfer weight of each type of working parameter in the parameter influence chain in step 104 above can be specifically implemented through the following steps: Step 301: Determine the superior working parameter of the target working parameter in the parameter influence chain.

[0056] Taking the parameter influence chain as “electricity-heat-pressure-humidity” as an example, if the target working parameter is heat, then its superior working parameter is electricity, and its subordinate working parameter is pressure.

[0057] Step 302: Based on the differences in real-time monitoring values ​​and out-of-class distances of different connectors under the upper-level working parameters, calculate the upper-level difference coefficient between any two connectors.

[0058] As one possible implementation, this step can be specifically implemented as follows: determining the absolute value of the difference between the real-time monitoring values ​​of the first connector and the second connector in the upper-level operating parameters; the first connector is any one of the multiple connectors, and the second connector is the connector other than the first connector among the multiple connectors; calculating the out-of-class distance between the first connector and the second connector under the upper-level operating parameters; and using the ratio of the absolute value to the out-of-class distance as the upper-level difference coefficient between the first connector and the second connector under the upper-level operating parameters.

[0059] As an example, the superior difference coefficient of the m-th type parameter between the k-th connector and the k'-th connector at time t. Satisfy the following formula: in, This represents the actual value monitored for the parent parameter of the k-th connector in the m-th parameter class. This represents the actual value monitored for the parent parameter of the k'-th connector in the m-th parameter class. This represents the out-of-class distance between the k-th connector and the k'-th connector in the m-th class of parameters, relative to their parent parameter. A small out-of-class distance coupled with a large difference in the real-time parent parameter indicates that the state difference between the two connectors is primarily due to the influence of that parent parameter. Therefore, the parent parameter difference coefficient... The larger. For parameter tuning coefficients, if If it is 0, then set it to 0.01. If it is not 0, then set it to 0.

[0060] Step 303: Determine the upper-level transfer coefficient based on the upper-level difference coefficient, the real-time monitoring value of the connector on the target working parameters and the deviation from the standard value.

[0061] It should be noted that the more concentrated the influence from the superior and the more the m-th type of parameter deviates from the set value, the greater the influence from the superior on the m-th type of parameter.

[0062] Optionally, the superior transmission coefficient is the product of the superior difference coefficient and the first absolute value, where the first absolute value is the absolute value of the difference between the real-time monitoring value of the target working parameter and the standard value.

[0063] As an example, the superior transfer coefficients of the k-th connector and the k'-th connector at time t under m types of parameters. Satisfy the following formula: in, This represents the superior difference coefficient between the m-th type parameter of the k-th connector and the k'-th connector at time t. This represents the real-time monitoring value of the m-th parameter of the k-th connector at time t. This represents the set standard value for the k-th connector in parameter m. This set standard value is the initial or target value of the parameter pre-set for each connector based on the test design and stress conditions. It is typically determined based on the test purpose, industry standards, or physical models (such as the Arrhenius model) to ensure that the test effectively accelerates the aging process. This represents the deviation between the actual value and the standard value of the m-th parameter of the k-th connector at time t. The larger this value is, the stronger the influence of the upper level on the actual value of the m-th parameter due to the difference in the upper level coefficient, and the larger the corresponding upper level transmission coefficient is.

[0064] Step 304: Based on the upper-level transfer coefficients of multiple connectors and the difference in the rate of change of performance degradation rate before and after the transfer moment, determine the transfer weight of the target operating parameters.

[0065] As one possible implementation, this step can be specifically implemented as follows: calculate the superior transfer coefficient between the first connector and each of the second connectors, and determine the maximum superior transfer coefficient; determine the time corresponding to the maximum superior transfer coefficient as the transfer time of the first connector; calculate the first difference between the average degradation rate change rate of the first connector after the transfer time and the first difference between the average degradation rate change rate before the transfer time; for each connector, calculate the product of the first difference and the corresponding maximum superior transfer coefficient; take the average of all the products and perform linear normalization to obtain the transfer weight of the target operating parameter. In other words, compare the superior transfer coefficients of the k-th connector with those of all other connectors at each time for the m-th type of parameter, and define the time corresponding to the maximum value as the transfer time of the m-th type of parameter of that connector. Based on the asymptotic nature of parameter influence, selecting the maximum value can effectively characterize the peak intensity of the parameter's influence from the superior. This transfer time marks the stage where the k-th connector is most significantly affected by the superior parameter for the m-th type of parameter, and the corresponding superior transfer coefficient at this time quantifies the maximum degree of superior influence on that connector.

[0066] As an example, the transitive weights of the m-th class of parameters Satisfy the following formula: in, This represents the total number of all connectors, where k is an integer greater than or equal to 1. This represents the parent-level transmission coefficient of the k-th connector at the transmission time for the m-th type of parameter. This represents the lifespan of the k-th connector. This indicates the time when the k-th connector transmits parameters of type m. Since the time when the working parameters are transmitted corresponds to the time of the maximum transmission coefficient of the superior parameter, the connector must be in normal working condition at this time. Therefore, the connector's lifespan must not have reached its maximum value at the time of transmission. The value of is always greater than , This represents the degradation rate of the k-th connector at the moment of transmission of its m-th type of parameter. This represents the rate of change of the average degradation rate of the k-th connector after the transmission time of the m-th type of parameter. A degradation rate of 1 indicates that the product has reached the failure threshold. This represents the rate of change of the average degradation rate of the k-th connector before the transmission time of the m-th type of parameter. This refers to the difference in degradation rate between the k-th connector and the point before and after the m-th parameter is most affected by the superior (i.e., the first difference mentioned above). The larger the difference, the more severe the accelerated degradation caused by the superior influence on the m-th parameter. This represents the linear normalization function.

[0067] It should be noted that if a target operating parameter does not have a parent operating parameter, it means that the target operating parameter is not affected by the parent, and the transfer weight of the target operating parameter can be directly set to 0. For example, in the parameter influence chain "electricity-heat-pressure-humidity", if the electricity operating parameter does not have a parent operating parameter, then the transfer weight of the electricity operating parameter is directly set to the preset value of 0.

[0068] Based on the above technical solution, this application constructs a complete chain for calculating the transfer weight, from identifying differences at the upper level to calculating the transfer coefficient, and finally dynamically determining the transfer weight in conjunction with performance degradation. This achieves a complete quantification of the cascading effects between parameters, providing a systematic solution for analyzing indirect effects under complex working conditions and enhancing the depth and dimension of reliability analysis.

[0069] like Figure 4 As shown, in one possible implementation, the process of determining the influence coefficient of each type of working parameter based on the basic relevance and transit weight of each type of working parameter in step 105 above can be specifically implemented through the following steps: Step 401: Perform a fusion calculation on the basic relevance of the target working parameters, the transfer weight of the target working parameters, and the transfer weight of the lower-level working parameters of the target working parameters.

[0070] Step 402: Normalize the fusion calculation results to obtain the influence coefficients of the target working parameters.

[0071] Among them, the influence coefficient is positively correlated with the basic correlation of the target working parameters and the transfer weight of the subordinate working parameters, and negatively correlated with the transfer weight of the target working parameters.

[0072] As an example, the influence coefficient of the m-th type parameter Satisfy the following formula: in, This represents the transit weight of the m-th type of parameter. This represents the transfer weight where the m-th type of parameter is used as the working parameter of the higher level. This represents the basic relevance of the m-th type of parameter. It should be noted that if the m-th type of working parameter does not have any subordinate working parameters (for example, the 4th-level working parameter "humidity" mentioned above does not have any subordinate parameters), it means that this working parameter has no subordinate transmission, and its corresponding parameter is directly assigned. Set to 0.

[0073] Based on the above technical solution, this application calculates the influence coefficient by integrating the basic correlation, its own transmission weight, and the lower-level transmission weight, and constructs an evaluation system that comprehensively considers the parameter's own influence, controllability, and dominance. This enables the influence coefficient to fully reflect the direct and indirect influence of the parameter on lifespan, and achieves a multi-dimensional and accurate assessment of the parameter's importance.

[0074] It should be noted that the order of the embodiments described above is merely for descriptive purposes and does not represent the superiority or inferiority of the embodiments. The processes depicted in the accompanying drawings do not necessarily require a specific or sequential order to achieve the desired result. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0075] The various embodiments in this specification are described in a progressive manner. The same or similar parts between the various embodiments can be referred to each other. Each embodiment focuses on describing the differences from other embodiments.

Claims

1. A connector reliability verification method that fuses artificial intelligence and complex environment simulation, characterized by, The method comprises: obtaining multiple types of working parameter data and life data of multiple connectors during an accelerated life test process; determining a basic correlation degree of each type of working parameter, the basic correlation degree being used to represent a degree of direct association between the working parameter and the life; constructing a parameter influence chain based on an influence transmission relationship between the multiple types of working parameters; determining a transmission weight of the each type of working parameter in the parameter influence chain, the transmission weight being used to represent a degree of indirect influence of the working parameter on the life caused by an influence of a superior working parameter; determining an influence coefficient of the each type of working parameter based on the basic correlation degree and the transmission weight of the each type of working parameter; based on the influence coefficient, constructing a connector reliability evaluation model through data fusion simulation and artificial intelligence model training, the connector reliability evaluation model being used to predict the life of the connector based on the influence coefficient of the connector.

2. The connector reliability verification method of claim 1, wherein The method comprises: calculating an out-of-class coefficient of each connector under a first working parameter, the out-of-class coefficient being used to represent a degree of overall similarity of the corresponding connector to other connectors under a second working parameter; the second working parameter being a working parameter other than the first working parameter; performing a weighted linear fitting on the first working parameter and the life data with the out-of-class coefficient of each connector as a weight; taking an absolute value of a slope of a straight line obtained through the fitting as the basic correlation degree of the first working parameter.

3. The hybrid artificial intelligence and complex environment simulation connector reliability verification method of claim 2, wherein, The method comprises: calculating a difference between the target connector and any other connector under all second working parameters as an out-of-class distance, the difference being a ratio of an absolute value of the difference between the target connector and any other connector under all second working parameters to a sample mean of the second working parameters; taking a mean of all out-of-class distances of the target connector as a degree of overall difference of the target connector; determining the out-of-class coefficient based on the degree of overall difference, the out-of-class coefficient being inversely proportional to the degree of overall difference.

4. The hybrid artificial intelligence and complex environment simulation connector reliability verification method of claim 1, wherein, The method comprises: determining a superior working parameter of a target working parameter in the parameter influence chain; calculating a superior difference coefficient between any two connectors based on a difference between real-time monitoring values of different connectors under the superior working parameter and an out-of-class distance; determining a superior transmission coefficient based on the superior difference coefficient and a deviation between real-time monitoring values of the connectors under the target working parameter and a standard value; determining a transmission weight of the target working parameter based on the superior transmission coefficients of the multiple connectors and a difference between performance degradation rate change speeds before and after a transmission time.

5. The hybrid artificial intelligence and complex environment simulation connector reliability verification method of claim 4, wherein, The method comprises: determining an absolute value of a difference between real-time monitoring values of a first connector and a second connector under the superior working parameter; the first connector being any connector in the multiple connectors, and the second connector being a connector other than the first connector in the multiple connectors; calculating out-of-class distances of the first connector and the second connector under the upper-level working parameter; calculating a ratio of the absolute value and the out-of-class distance as an upper-level difference coefficient of the first connector and the second connector under the upper-level working parameter.

6. The hybrid artificial intelligence and complex environment simulation connector reliability verification method of claim 4, wherein, The upper-level transfer coefficient is a product of the upper-level difference coefficient and a first absolute value, and the first absolute value is an absolute value of a difference between a real-time monitoring value of the target working parameter and a standard value.

7. The hybrid artificial intelligence and complex environment simulation connector reliability verification method of claim 4, wherein, Based on the upper-level transfer coefficients of the plurality of connectors and a difference in performance degradation rate change speed before and after a transfer time, a transfer weight of the target working parameter is determined, including: calculating upper-level transfer coefficients between the first connector and each second connector, and determining a maximum upper-level transfer coefficient among them; determining a time corresponding to the maximum upper-level transfer coefficient as the transfer time of the first connector; calculating a first difference value of an average degradation rate change speed of the first connector after the transfer time and an average degradation rate change speed before the transfer time; for each connector, calculating a product of the first difference value of the connector and the corresponding maximum upper-level transfer coefficient; and performing linear normalization on an average value of all the products to obtain the transfer weight of the target working parameter.

8. The hybrid artificial intelligence and complex environment simulation connector reliability verification method of claim 1, wherein, Based on the basic correlation degree of each type of working parameter and the transfer weight, an influence coefficient of each type of working parameter is determined, including: performing fusion calculation and normalization on the basic correlation degree of the target working parameter, the transfer weight of the target working parameter, and the transfer weight of the lower-level working parameter of the target working parameter to obtain the influence coefficient of the target working parameter; the influence coefficient is positively correlated with the basic correlation degree of the target working parameter and the transfer weight of the lower-level working parameter, and is negatively correlated with the transfer weight of the target working parameter.

9. The hybrid artificial intelligence and complex environment simulation connector reliability verification method of claim 1, wherein, Based on the influence coefficient, a connector reliability evaluation model is constructed through data fusion simulation and artificial intelligence model training, including: using a working parameter with an influence coefficient higher than a preset threshold as an input variable to establish a parameterized finite element model; performing virtual accelerated life testing in a simulation environment to generate a simulation data set including parameter evolution and virtual life; inputting the simulation data set into a data generation model to generate virtual sample data; the data generation model is a pre-trained neural network model for data generation; generating a training set based on actual test data, the simulation data set, and the virtual sample data, and training a physical information neural network to obtain the connector reliability evaluation model.

10. The hybrid artificial intelligence and complex environment simulation connector reliability verification method of claim 1, wherein, Obtaining a plurality of working parameter data and life data of a plurality of connectors during accelerated life testing, including: randomly assigning the plurality of connectors to different stress groups for accelerated life testing, the stress groups including temperature stress groups and humidity stress groups; the plurality of working parameters include temperature parameters and humidity parameters; periodically monitoring the environmental parameters and insulation resistance of each connector during testing; determining the life data of each connector according to the time when the insulation resistance drops to a preset percentage of the initial value.

Images